Prediction of wind farm reactive power fast variations by adaptive one-dimensional convolutional neural network. (December 2021)

• Record Type: Journal Article
• Authors: Samet, Haidar; Ketabipour, Saeedeh; Afrasiabi, Shahabodin; Afrasiabi, Mousa; Mohammadi, Mohammad
• Is Part Of: Computers & electrical engineering. Volume 96:Part A(2021)
• Publication Date: 2021-12
• Subjects: Wind energy -- Flicker emission -- Deep neural network -- One dimensional convolutional deep learning network -- Static VAr compensator -- Wind farm; Computer engineering -- Periodicals; Electrical engineering -- Periodicals; Electrical engineering -- Data processing -- Periodicals; Ordinateurs -- Conception et construction -- Périodiques; Électrotechnique -- Périodiques; Électrotechnique -- Informatique -- Périodiques; Computer engineering; Electrical engineering; Electrical engineering -- Data processing; Periodicals; Electronic journals; 621.302854
• DOI: 10.1016/j.compeleceng.2021.107480
• Languages: English
• ISSNs: 0045-7906

Abstract

Abstract: One of the prominent problems in wind farms is voltage flicker emission. To prevent flicker emission or mitigate the impact as best as possible, a static VAr compensator (SVC) is a great candidate both economically and technically. However, SVCs cannot completely compensate the fast-changing reactive power due to delays caused by the reactive power calculation unit and the triggering fire angle of the SVC. This paper proposes a predictive control system for SVCs, by merging an additional predictive control block into the conventional control system. It is constructed based on deep neural networks, namely adaptive one-dimensional convolutional neural network (1D-CNN). The training process is conducted based on the adaptive learning weights process to enhance the prediction accuracy and training computational complexity of the 1D-CNN. Numerical results on the actual dataset in a wind farm in Manjil, Iran, have verified the forecasting accuracy and flicker mitigation of the proposed controller.